Evaluating computational models of explanation using human judgments

Michael Pacer, Tania Lombrozo, Thomas Griffiths, Joseph Williams, Xi Chen
Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, PMLR R11:193-202, 2013.

Abstract

We evaluate four computational models of ex- planation in Bayesian networks by compar- ing model predictions to human judgments. In two experiments, we present human par- ticipants with causal structures for which the models make divergent predictions and either solicit the best explanation for an observed event (Experiment 1) or have participants rate provided explanations for an observed event (Experiment 2). Across two versions of two causal structures and across both exper- iments, we find that the Causal Explanation Tree and Most Relevant Explanation mod- els provide better fits to human data than either Most Probable Explanation or Expla- nation Tree models. We identify strengths and shortcomings of these models and what they can reveal about human explanation. We conclude by suggesting the value of pur- suing computational and psychological inves- tigations of explanation in parallel.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR11-pacer13a, title = {Evaluating computational models of explanation using human judgments}, author = {Pacer, Michael and Lombrozo, Tania and Griffiths, Thomas and Williams, Joseph and Chen, Xi}, booktitle = {Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence}, pages = {193--202}, year = {2013}, editor = {Nicholson, Ann and Smyth, Padhraic}, volume = {R11}, series = {Proceedings of Machine Learning Research}, month = {12--14 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r11/main/assets/pacer13a/pacer13a.pdf}, url = {https://proceedings.mlr.press/r11/pacer13a.html}, abstract = {We evaluate four computational models of ex- planation in Bayesian networks by compar- ing model predictions to human judgments. In two experiments, we present human par- ticipants with causal structures for which the models make divergent predictions and either solicit the best explanation for an observed event (Experiment 1) or have participants rate provided explanations for an observed event (Experiment 2). Across two versions of two causal structures and across both exper- iments, we find that the Causal Explanation Tree and Most Relevant Explanation mod- els provide better fits to human data than either Most Probable Explanation or Expla- nation Tree models. We identify strengths and shortcomings of these models and what they can reveal about human explanation. We conclude by suggesting the value of pur- suing computational and psychological inves- tigations of explanation in parallel.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Evaluating computational models of explanation using human judgments %A Michael Pacer %A Tania Lombrozo %A Thomas Griffiths %A Joseph Williams %A Xi Chen %B Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2013 %E Ann Nicholson %E Padhraic Smyth %F pmlr-vR11-pacer13a %I PMLR %P 193--202 %U https://proceedings.mlr.press/r11/pacer13a.html %V R11 %X We evaluate four computational models of ex- planation in Bayesian networks by compar- ing model predictions to human judgments. In two experiments, we present human par- ticipants with causal structures for which the models make divergent predictions and either solicit the best explanation for an observed event (Experiment 1) or have participants rate provided explanations for an observed event (Experiment 2). Across two versions of two causal structures and across both exper- iments, we find that the Causal Explanation Tree and Most Relevant Explanation mod- els provide better fits to human data than either Most Probable Explanation or Expla- nation Tree models. We identify strengths and shortcomings of these models and what they can reveal about human explanation. We conclude by suggesting the value of pur- suing computational and psychological inves- tigations of explanation in parallel. %Z Reissued by PMLR on 04 October 2026.
APA
Pacer, M., Lombrozo, T., Griffiths, T., Williams, J. & Chen, X.. (2013). Evaluating computational models of explanation using human judgments. Proceedings of the 29th Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R11:193-202 Available from https://proceedings.mlr.press/r11/pacer13a.html. Reissued by PMLR on 04 October 2026.

Related Material